deviation of 188 psi were astounding. The difference between FEM static 

 stresses and prototype data is thought to be due to the wedging loads re- 

 sulting from the binding effect of the nesting armor layer. The prototype 

 data also showed the maximum pulsating stress to be surprisingly well-behaved 

 and highly correlated with the average of the highest one -tenth of the waves 

 (H 1/10 ) (Howell, Rhee , and Rosati 1990). Additionally, the prototype data 

 revealed that the dolos static stress was increasing over time at a rate of 

 approximately 26 psi per year (Melby and Howell 1989, Kendall and Melby 1992). 



30. While the Crescent City prototype data yielded considerable in- 

 sight into the dolos structural response, it was also used in the physical 

 model dolos instrumentation verification. Markle (1990a) showed that small- 

 scale dolosse embedded with the load cell could be used to measure pulsating 

 stresses in the physical model that could be correctly scaled to prototype. 

 Thus, structural measurements could be done simultaneously with hydrodynamic 

 stability studies. Burcharth and Liu (1990) later showed that the load cell 

 could also be used to measure scalable static response in the physical model, 

 provided the dolos surface friction was correctly scaled. The load cell has 

 not been calibrated for impact response. Because the material elasticity 

 scaling is difficult with the load cell, accurate measurement of impact 

 response in the small-scale model using this instrument requires more 

 research. 



31. As stated earlier, the Crescent City prototype data were augmented 

 by deterministic results from FEM studies to arrive at the dolos design pro- 

 cedure as presented herein (Melby and Howell 1989) . This design procedure 

 utilizes FEM results to modify scaled prototype stress distributions. Static 

 and pulsating stress distributions are modified for waist ratio, stacking 

 depth, material unit weight, and dolos size. The modified distributions are 

 combined to achieve a design stress distribution. So, given a design stress 

 probability, the designer can determine a design stress level for the armor 

 layer. The design stress is compared to a fatigue -reduced strength to give a 

 factor of safety. All of these calculations are incorporated into the micro- 

 computer program CAUDAID. 



17 



